Crossover Trials in Diabetes Research

Crossover designs are efficient alternatives to parallel designs. Sample size reduction, comparing the 2x2 crossover to a two-arm parallel study, is approximately 4 to 1 in diabetes clinical trials. One challenge of using a crossover design is understanding and handling potential carryover effects. This paper will review the fundamentals of the crossover design, including design, model, and analyses. The problems of the carryover effect will be addressed. We discuss Grizzle’s [Grizzle(1965)] well accepted two-stage approach. This approach has serious flaws that we will summarize as well as other views of handling the carryover effect from [Freeman(1989)], [Lehmacher(1991)] and [Senn(1993), Senn(1997)]. Lastly, we assess the influence of carryover effect from results of several clinical trials to evaluate the legitimacy of the use of crossover designs in diabetes research. We will also give a recommendation for the conduct and analysis of crossover trials.